We are developing means to reduce errors in Dipole Source Localization (DSL). DSL is commonly used to analyze brain waves for clinical and research purposes. It is known from our work and the work of others that the models used to compute the DSL are not perfect, but it is not known how much the imperfect models affect the solutions obtained. Based upon our prior work, and some new concepts, we have developed the DSME (Dipole-Source Model- Error). We have tested its efficacy in spheres, and will do so in headshape models in this project. We will compare errors both by simulations, and also by determining whether the DSL has errors when analyzing evoked-responses from known neural generators. The health-related aspects include the clinical uses of evoked-responses and DSL, and the fact that improvements in the accuracy of the technique will improve these uses.